Abstract
Since the launch of ChatGPT in Dec 2022, Large Language Models (LLMs) are rapidly being adopted by businesses to assist users in a wide range of open-ended tasks, including ones that require creativity. We compare human-LLM collaboration where the model acts as a ghostwriter versus a sounding board and show that the collaboration modality changes outcomes meaningfully across expertise levels.
Joint work with Jason Chan.
This paper studies how different human-LLM collaboration modalities shape creative work outcomes.